Security audit

Local Search

Security checks across malware telemetry and agentic risk

Overview

This is a coherent local web-search skill that sends searches to public search engines, with no hidden persistence, credential access, or unrelated behavior found.

Before installing, treat searches as network requests to public search providers: your query, IP address, locale-like signals, user-agent, and timing may be visible to Brave, DuckDuckGo, Bing, or Google depending on engine selection and fallback. Avoid sending secrets or highly sensitive queries, and use the explicit output option only for files where you intend to save results.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (5)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 94% confidence
Finding: The skill clearly performs outbound network access to third-party search engines, but the metadata shown in this file does not declare permissions. Undeclared network capability reduces transparency and weakens policy enforcement, making it easier for users or agents to invoke data-exfiltrating behavior without an explicit trust decision.

Missing User Warnings

Low

Confidence: 81% confidence
Finding: The README explicitly promotes direct HTTP scraping of external search engines from the user's machine, but does not clearly warn that user queries, IP address, user-agent, locale, and timing metadata will be transmitted to third-party search providers. This is a genuine privacy/security disclosure gap: users may assume 'local' means fully offline or non-transmitting, when the skill in fact sends sensitive search terms over the network.

Missing User Warnings

Medium

Confidence: 90% confidence
Finding: The skill sends user queries directly to third-party search engines, which necessarily exposes search terms and the client's IP address to external services. While the document later mentions privacy, the normal-use disclosure is not prominent enough near the description and usage path, so users may invoke it without understanding the privacy consequence.

Missing User Warnings

Low

Confidence: 91% confidence
Finding: This code sends the user's search query to DuckDuckGo over the network, which exposes potentially sensitive user input to a third-party search provider. In the context of a local-search skill advertised as running on the user's machine with direct scraping and no cloud hop, the risk is primarily privacy-related rather than code-execution, but lack of explicit user disclosure or consent can still leak confidential queries.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: This code sends the user's query and inferred or explicit locale to third-party search engines via `callEngine(...)`, which can expose sensitive user intent, personal data, or location/language cues to external services. In the context of a 'local' search skill that emphasizes running on the user's machine and no cloud hop, the lack of an explicit disclosure/consent mechanism in this path increases privacy risk because users may incorrectly assume their data stays local.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.